qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [RFC PATCH 2/2] GitLab CI: crude mapping of PMM's scripts to jobs


From: Cleber Rosa
Subject: Re: [RFC PATCH 2/2] GitLab CI: crude mapping of PMM's scripts to jobs
Date: Tue, 10 Mar 2020 01:01:06 -0400

On Sat, Feb 08, 2020 at 01:02:41PM +0000, Peter Maydell wrote:
> On Fri, 7 Feb 2020 at 19:34, Cleber Rosa <address@hidden> wrote:
> > Also, the build and make steps worked fine without `--disable-gnutls`.
> > I was trying to not carry over any "exception" type of arguments,
> > unless they proved to be necessary.  Maybe Peter can give some more
> > info about this (should it be kept or not)?
> 
> A lot of those config options are random historical accident:
> I probably added in the configure option to work around something
> years back and then never bothered to undo it. I agree that if we
> don't need to pass that option to run on whatever OS the CI
> job is running on we shouldn't keep it.
> 
> More generally, I don't think we should worry much about
> exactly replicating the fine detail of the various configurations I

OK, good to hear that.

> currently run on x86. If we basically cover:
>  * a debug build

Which should be covered by the following job (relevant snippets only):

ubuntu-18.04.3-x86_64-alldbg:
 ...
 - ./configure --enable-debug --disable-libssh
 - make clean
 - make
 - make check V=1

>  * a non-debug build

I assume the following is a suitable non-debug build:

ubuntu-18.04.3-x86_64-all:
 ...
 script:
 - ./configure --disable-libssh
 - make
 - make check V=1

>  * a linux-user --static build

This is roughly the job to fulfill this requirement:

ubuntu-18.04.3-x86_64-all-linux-static:
 ...
 # --disable-libssh is needed because of 
https://bugs.launchpad.net/qemu/+bug/1838763
 # --disable-glusterfs is needed because there's no static version of those 
libs in distro supplied packages
 - ./configure --enable-debug --static --disable-system --disable-glusterfs 
--disable-libssh
 - make
 - make check V=1
 - make check-tcg V=1

>  * a clang build with the sanitizers enabled [+]

I've tested the following which includes the sanitizers but drops the
'--enable-gtk' option:

ubuntu-18.04.3-x86_64-clang:
 ...
 - ./configure --disable-libssh --cc=clang --cxx=clang++ --enable-sanitizers
 - make
 - make check V=1

>  * windows crossbuilds

OK, there weren't any windows crossbuilds jobs in this version but I'm
adding them (32 and 64 bits).

>  * a --disable-tcg build

Which should be covered by the following job definition:

ubuntu-18.04.3-x86_64-notcg:
 ...
 - ./configure --disable-libssh --disable-tcg
 - make
 - make check V=1

>  * an --enable-tci-interpreter build

Which should be covered by:

ubuntu-18.04.3-x86_64-tci:
 ...
 - ./configure --disable-libssh --enable-tcg-interpreter
 - make

>  * and at least one of the above is done as an "incremental"
>    build and one as a "make clean and then build"

I'm not sure yet how to cleanly accomplish this.  IIRC it suggests
working with different versions of the code in a single job.
GitLab will usually give you the code matching the commit you're
testing.  I believe I could do something like:

 - git reset --hard HEAD~X
 - configure ...
 - make
 - git checkout $PREVIOUS_HEAD
 - make

But I'll have to experiment a bit more with it.

> then that's the same coverage we have today.
> 
> [+] my scripts do this by hand by passing a lot of extra cflags,
> but IIRC configure now supports a simple 'enable sanitizers'
> option of some kind, which would be OK too
> 
> thanks
> -- PMM
> 

Thanks for the very straight answers here.  I'll work
on those two remaining points (windows cross builds, and
incremental builds).

Please let me know if I misunderstood any of your points.

Thanks,
- Cleber.

Attachment: signature.asc
Description: PGP signature


reply via email to

[Prev in Thread] Current Thread [Next in Thread]